CN118152836B - Stability evaluation method for operation process of electric energy meter - Google Patents

Abstract

The invention relates to the technical field of data processing, in particular to a stability evaluation method for an operation process of an electric energy meter, which is used for collecting current time sequence data in a preset period in the operation process of the electric energy meter, optimizing the corresponding clustering distance of two data points in a clustering process aiming at any two data points in the current time sequence data, and correspondingly obtaining the final clustering distance between the two data points; based on the final clustering distance between every two data points in the current time sequence data, obtaining a corresponding clustering result, and removing a high-load fluctuation clustering cluster from the clustering result to obtain a final clustering result; according to the clustering discrete features in the final clustering result, determining the stability evaluation result of the electric energy meter in a preset period, so that the high-load mutation data is prevented from being divided as abnormal data, and the accuracy of the stability evaluation result of the electric energy meter according to the clustering result is improved.

Description

Stability evaluation method for operation process of electric energy meter

Technical Field

The invention relates to the technical field of data processing, in particular to a stability evaluation method for an electric energy meter operation process.

Background

The electric energy meter is an instrument for measuring electric energy consumption and is widely applied to the fields of industrial and commercial electricity metering, energy monitoring, charging and the like. In the use process of the electric energy meter, the precision and the stability of the electric energy meter are two important performance indexes. Because factors such as temperature change, humidity change and long-term use abrasion influence the stability of the electric energy meter, the stability evaluation of the electric energy meter is helpful for timely finding and solving potential problems.

In the prior art, when stability evaluation is performed on the electric energy meter, current time sequence data of the electric energy meter is generally clustered through a K-means clustering algorithm, one cluster represents an electricity consumption mode, and further, the stability of the electric energy meter can be evaluated according to data discrete conditions in the clustered clusters and error requirements of the electric energy meter. However, the current time sequence data collected by the electric energy meter not only has current data under normal operation, but also has current data fluctuation caused when high-load equipment is suddenly connected or disconnected, and the current data fluctuation under the conditions is irrelevant to the operation stability of the electric energy meter, but in the clustering analysis process, the current data under the conditions can be considered as fluctuation data under the condition of unstable operation of the electric energy meter, so that a clustering result is influenced, and the stability assessment of the electric energy meter is inaccurate.

Therefore, how to reduce the influence of high-load mutation on the clustering result so as to improve the accuracy of the stability evaluation of the electric energy meter according to the clustering result is a problem to be solved.

Disclosure of Invention

In view of the above, the embodiment of the invention provides a stability evaluation method for an operation process of an electric energy meter, so as to solve the problem of how to reduce the influence of high-load mutation on a clustering result and improve the accuracy of stability evaluation of the electric energy meter according to the clustering result.

The embodiment of the invention provides a stability evaluation method for an operation process of an electric energy meter, which comprises the following steps:

during the operation process of the electric energy meter, collecting current time sequence data in a preset period, and taking any data point in the current time sequence data as a center to obtain a local data segment of the data point;

for any two data points in the current time sequence data, according to the similarity between the local data segments of the two data points, carrying out initial optimization on the clustering distance corresponding to the two data points in the clustering process, and correspondingly obtaining the clustering optimization distance between the two data points;

for any one of the two data points, acquiring a distance optimization factor of the data point according to the data fluctuation of the local data segment of the data point, and re-optimizing the clustering optimization distance between the two data points according to the distance optimization factor of the two data points to correspondingly obtain the final clustering distance between the two data points;

Based on the final clustering distance between every two data points in the current time sequence data, clustering all data points in the current time sequence data to obtain a corresponding clustering result, and removing a high-load fluctuation clustering cluster from the clustering result to obtain a final clustering result;

And determining a stability evaluation result of the electric energy meter in the preset period according to the clustering cluster discrete features in the final clustering result.

Preferably, the performing initial optimization on the clustering distance corresponding to the two data points in the clustering process according to the similarity between the local data segments of the two data points, and correspondingly obtaining the clustering optimization distance between the two data points includes:

Acquiring a DTW distance between local data segments of the two data points, and carrying out normalization processing on the DTW distance, wherein an obtained normalization processing result is used as a time sequence influence factor;

And obtaining the corresponding clustering distance of the two data points in the clustering process, obtaining the addition result between a constant 1 and the time sequence influence factor, and taking the product between the addition result and the clustering distance as the clustering optimization distance between the two data points.

Preferably, the obtaining the distance optimization factor of the data point according to the data fluctuation of the local data segment of the data point includes:

Dividing the difference value between the maximum value and the minimum value in the local data segment of the data point into a preset number of sections according to the dividing result, taking the ratio between the number of the data points contained in each section and the total number of the data points in the local data segment as the probability of the corresponding section respectively, acquiring the corresponding entropy value according to the probability of each section, and acquiring the range and the data value average value in the local data segment of the data point to obtain the first ratio between the range and the data value average value;

Calculating the absolute value of a difference between two adjacent data points from the first data point of the local data section of the data point, recording a first time stamp of the first data point if the absolute value of the difference is larger than a preset difference threshold value, continuing traversing the data points in the local data section of the data point until the absolute value of the difference is smaller than or equal to the difference threshold value, recording a second time stamp of the absolute value of the difference which is larger than the difference threshold value for the last time, acquiring the time interval between the first time stamp and the second time stamp, and taking the addition result of all the time intervals as fluctuation duration time if at least two time intervals exist in the local data section;

And acquiring the duty ratio of the fluctuation duration in the duration corresponding to the local data segment of the data point, substituting the product among the entropy value, the first ratio and the duty ratio into an exponential function with a natural constant as a base, and taking the obtained function result as a distance optimization factor of the data point.

Preferably, the re-optimizing the cluster optimization distance between the two data points according to the distance optimization factor of the two data points, correspondingly obtaining the final cluster distance between the two data points, includes:

And taking the multiplication result of the distance optimization factor of the two data points and the clustering optimization distance between the two data points as the final clustering distance between the two data points.

Preferably, the removing the high load fluctuation cluster from the cluster result to obtain a final cluster result includes:

according to the data average value of each cluster in the clustering result, obtaining the average value of the data average values of all clusters, and determining that the clusters are high-load fluctuation clusters according to any cluster in the clustering result if the data average value of the clusters is larger than the average value of a first preset proportion or smaller than the average value of a second preset proportion;

Traversing each cluster in the clustering result to obtain all high-load fluctuation clusters, and removing all the high-load fluctuation clusters in the clustering result to obtain a final clustering result.

Preferably, the determining, according to the cluster discrete feature in the final cluster result, a stability evaluation result of the electric energy meter within the preset period includes:

Respectively acquiring the data range and the data standard deviation of each cluster in the final clustering result to obtain the maximum data range and the maximum data standard deviation;

and if the maximum data range is smaller than or equal to the standard data range and the maximum data standard deviation is smaller than or equal to the standard data standard deviation, determining that the electric energy meter is in a stable running state in the preset period.

Preferably, the clustering, based on the final clustering distance between every two data points in the current time series data, clusters all data points in the current time series data to obtain a corresponding clustering result, including:

And acquiring a K value when K-means clustering is carried out on all data points in the current time sequence data through an elbow rule, and carrying out K-means clustering on all data points in the current time sequence data based on the K value and the final clustering distance between every two data points in the current time sequence data to obtain a corresponding clustering result.

Compared with the prior art, the embodiment of the invention has the beneficial effects that:

In the running process of the electric energy meter, current time sequence data in a preset period are collected, and a local data segment of any data point in the current time sequence data is obtained by taking the data point as the center; for any two data points in the current time sequence data, according to the similarity between the local data segments of the two data points, carrying out initial optimization on the clustering distance corresponding to the two data points in the clustering process, and correspondingly obtaining the clustering optimization distance between the two data points; for any one of the two data points, acquiring a distance optimization factor of the data point according to the data fluctuation of the local data segment of the data point, and re-optimizing the clustering optimization distance between the two data points according to the distance optimization factor of the two data points to correspondingly obtain the final clustering distance between the two data points; based on the final clustering distance between every two data points in the current time sequence data, clustering all data points in the current time sequence data to obtain a corresponding clustering result, and removing a high-load fluctuation clustering cluster from the clustering result to obtain a final clustering result; and determining a stability evaluation result of the electric energy meter in the preset period according to the clustering cluster discrete features in the final clustering result. The method comprises the steps of obtaining local data segments of each data point in current time sequence data, adding time sequence constraint when the current time sequence data are clustered, initially optimizing clustering distance among data points according to the local time sequence data among the data points, and meanwhile considering that partial fluctuation data mode features of an unstable electric energy meter are similar to high-load abrupt change data mode features in the clustering process, data under the conditions are easily clustered into one type, so that the clustering distance after the initial optimization of the added time sequence constraint is optimized again through the local fluctuation features of the local data segments of each data point in the current time sequence data, the clustering distance between any two data points in the clustering process is more rigorous, the high-load abrupt change data is prevented from being divided as abnormal data, and accuracy of stability evaluation results of the electric energy meter according to clustering results is improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the embodiments or the description of the prior art will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a flowchart of a method for evaluating stability of an operation process of an electric energy meter according to an embodiment of the present invention.

Detailed Description

Embodiments of the present disclosure are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and intended for the purpose of explaining the present disclosure and are not to be construed as limiting the present disclosure.

It should be noted that the terms "first," "second," and the like in the description of the present disclosure and the above-described figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the disclosure described herein may be capable of operation in sequences other than those illustrated or described herein. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with aspects of the present disclosure.

In order to illustrate the technical scheme of the invention, the following description is made by specific examples.

Referring to fig. 1, a method flowchart of a method for evaluating stability of an operation process of an electric energy meter according to an embodiment of the present invention is shown in fig. 1, where the method may include:

Step S101, collecting current time sequence data in a preset period in the operation process of the electric energy meter, and taking any data point in the current time sequence data as a center to obtain a local data segment of the data point.

The current sensor inside the electric energy meter can be used for collecting current data in the operation process of the electric energy meter, so that the current time sequence data in a preset period are collected in the operation process of the electric energy meter, and each time stamp corresponds to one data point for the collected current time sequence data. Preferably, the preset period may be 1 hour, 2 hours, one day, etc., without limitation.

After the current time sequence data is obtained, the current time sequence data is traditionally clustered by using a K-means clustering algorithm, one cluster represents an electricity utilization mode, so that a plurality of clusters are obtained, and the stability evaluation of the electric energy meter is carried out according to the cluster classification result. However, the operation process of the electric energy meter has normal electricity consumption data and also comprises the influence of high-load mutation data, the high-load mutation data can be forcedly divided into similar clusters in k-means clustering, and the stability analysis is performed on the subsequent cluster result to cause errors.

Before optimizing the clustering distance between every two data points in the clustering process, a window with the window size of 10 is set by taking any data point in the current time sequence data as the center, all the data points in the window form a local data segment of the data point, and the local data segment of each data point in the current time sequence data can be obtained in the same way.

Step S102, for any two data points in the current time sequence data, according to the similarity between the local data segments of the two data points, the clustering distance corresponding to the two data points in the clustering process is initially optimized, and the clustering optimization distance between the two data points is correspondingly obtained.

The time sequence distance between the local data segments of every two data points in the current time sequence data represents the similarity of the data fluctuation modes, when a high load mutation occurs, a certain data point in the collected current time sequence data can suddenly rise or fall in the moment, and the time sequence distance between the local data segment and the local data segment collected in normal operation is increased under the condition, which represents that the fluctuation modes corresponding to the data points are different. Calculation of the time sequence distance for all pairs of data points in the current time sequence data can measure the fluctuation mode characteristics represented by the window in which the selected data point is located. Therefore, for any two data points in the current time sequence data, according to the similarity between local data segments of the two data points, the clustering distance corresponding to the two data points in the clustering process is initially optimized, and the clustering optimization distance between the two data points is correspondingly obtained.

Preferably, according to the similarity between the local data segments of the two data points, the initial optimization is performed on the clustering distance corresponding to the two data points in the clustering process, so as to obtain the clustering optimization distance between the two data points correspondingly, which includes:

Acquiring a DTW distance between local data segments of the two data points, and carrying out normalization processing on the DTW distance, wherein an obtained normalization processing result is used as a time sequence influence factor;

And obtaining the corresponding clustering distance of the two data points in the clustering process, obtaining the addition result between a constant 1 and the time sequence influence factor, and taking the product between the addition result and the clustering distance as the clustering optimization distance between the two data points.

In one embodiment, taking a clustering distance between an ith data point and a jth data point in the current time series data as an example, a calculation expression of a time series influence factor between the ith data point and the jth data point is as follows:

Wherein, Represents the timing impact factor between the ith data point and the jth data point, norm () represents the normalization function,Represents the DTW distance between the local data segment of the i-th data point and the local data segment of the j-th data point,A local data segment representing the ith data point,Representing a local data segment for the jth data point.

It should be noted that, the acquisition of the DTW distance belongs to the prior art, and is not described here again; secondly, the larger the DTW distance between the local data segment of the i-th data point and the local data segment of the j-th data point, the larger the fluctuation mode difference corresponding to the two data points, and the larger the time sequence influence factor between the corresponding i-th data point and the j-th data point.

After the time sequence influence factor between the ith data point and the jth data point is obtained, the time sequence influence factor between the ith data point and the jth data point can be utilized to optimize the clustering distance between the ith data point and the jth data point in the clustering process, so as to obtain the clustering optimization distance, and the calculation expression of the clustering optimization distance is as follows:

Wherein, Representing the cluster optimization distance between the i-th data point and the j-th data point,Representing the euclidean distance between the ith and jth data points during clustering, 1 representing a constant,Representing the timing impact factor between the i-th data point and the j-th data point.

The larger the time sequence influence factor between the ith data point and the jth data point is, the larger the difference between the fluctuation mode features represented by the window where the data points are located is, the less likely the data points are classified into one class, and the clustering distance between the ith data point and the jth data point in the clustering process is increased, so that the larger the clustering optimization distance between the corresponding ith data point and the jth data point is.

Step S103, for any one of the two data points, according to the data fluctuation of the local data segment of the data point, obtaining the distance optimization factor of the data point, and according to the distance optimization factor of the two data points, re-optimizing the clustering optimization distance between the two data points, and correspondingly obtaining the final clustering distance between the two data points.

The high load mutation is considered to have no influence on the generation of the situation no matter whether the electric energy meter is stable or not, and the high load mutation is an external fluctuation which is irrelevant to the stability of the electric energy meter, has accidental occurrence and has similarity in data fluctuation modes. However, when the stability of the electric energy meter is poor, the data within a period of time have the same data fluctuation mode, the data fluctuation generated by the instability of the electric energy meter is similar to the data fluctuation generated by the high load mutation, and the data fluctuation is easily gathered into one type in the clustering analysis using the time sequence distance constraint, because in the clustering after the time sequence constraint, the fluctuation data are similar when the DTW distance is calculated, so that the high load mutation data are easily used as the fluctuation data (namely abnormal data) of the unstable electric energy meter, and the data are classified into one type, therefore, the high load mutation data need to be removed in the stability evaluation process of the electric energy meter based on the clustering result, so that the accuracy of the stability evaluation result of the electric energy meter is ensured, and the distance optimization factors of the ith data point and the jth data point in the current time sequence data are respectively obtained according to the data characteristics of the electric energy meter when the high load mutation occurs.

For any one of the two data points, according to the data fluctuation of the local data segment of the data point, acquiring a distance optimization factor of the data point, wherein the distance optimization factor acquiring method comprises the following steps:

Dividing the difference value between the maximum value and the minimum value in the local data segment of the data point into a preset number of sections according to the dividing result, taking the ratio between the number of the data points contained in each section and the total number of the data points in the local data segment as the probability of the corresponding section respectively, acquiring the corresponding entropy value according to the probability of each section, and acquiring the range and the data value average value in the local data segment of the data point to obtain the first ratio between the range and the data value average value;

Calculating the absolute value of a difference between two adjacent data points from the first data point of the local data section of the data point, recording a first time stamp of the first data point if the absolute value of the difference is larger than a preset difference threshold value, continuing traversing the data points in the local data section of the data point until the absolute value of the difference is smaller than or equal to the difference threshold value, recording a second time stamp of the absolute value of the difference which is larger than the difference threshold value for the last time, acquiring the time interval between the first time stamp and the second time stamp, and taking the addition result of all the time intervals as fluctuation duration time if at least two time intervals exist in the local data section;

And acquiring the duty ratio of the fluctuation duration in the duration corresponding to the local data segment of the data point, substituting the product among the entropy value, the first ratio and the duty ratio into an exponential function with a natural constant as a base, and taking the obtained function result as a distance optimization factor of the data point.

In one embodiment, when the high load device is suddenly turned on or off, the current data appearing at a single or several consecutive points in time in the current time series data rises or falls sharply, because an instantaneous rush current is generated at the time of turning on or off the high load device, which rises or falls sharply at the moment of turning on or off, and the current time series data of the electric energy meter is maintained within a stable range as the capacitor charging is completed and the magnetic field of the inductive element tends to settle sharply. Unlike the high load abrupt change, the fluctuation of the data generated by the poor stability of the electric energy meter can continuously exist, and is not generated or disappeared along with the change of the external load, and the peak-to-peak value is smaller than the value in the local window range of the high load abrupt change data point although the frequency and the amplitude are uncertain. Thus, the local data segment for the ith data pointIn the measuring ofAccording to the degree of fluctuation of the data of (1)The difference between the maximum and minimum of (2) bisects 10 segments, for example: assume that1, 1.1, 1.2, 2, 3, 4, 4.2, 5, 5.1, 11,The difference between the maximum value and the minimum value of (2) is 10, and the maximum value and the minimum value of (2) are calculated according to the difference of 10Divided into 10 sections, each section having a spacing of 1 corresponding to the first section of [1,2 ], [2,3 ], [3,4 ], [4,5 ], [5,6 ], [6,7 ], [7,8 ], [8,9 ], [9, 10 ], [10, 11], each section being marked with a corresponding level from low to highQ= [1,10], willDividing the data points in each section according to the range of each section so as to obtain the number of the data points contained in each section, and further dividing the number of the data points contained in each section with the local data sectionThe ratio between the total number of data points in (a) is taken as the probability of the corresponding segment, respectively, for example: the number of data points contained in the section [1,2 ] is 3, and the corresponding probability is: p=3/15=1/5.

After deriving the probability for each segment, calculating the entropy value as a local data segment of the ith data point based on the probability for each segmentThe calculation method of the volatility optimization factor and the entropy value of the model belongs to the prior art, and is not repeated here. Due to a single local data segment according to the ith data pointThe entropy change in the data wave mode can not be identified in specific high-load abrupt change, and the probability is not clear in the partial data segmentThe characteristics of the data pattern in the memory have the influence on the entropy value of the memory due to the existence of abnormal values, so that the analysis of the local data segment is also neededDuration of fluctuation in (a). Specifically, a difference threshold is set, which ranges from a local data segmentIt should be noted that, the data mean value of (1%) is set according to the minimum requirements of the stability standard of the electric energy meters of different types and grades in national standard, the value is the upper limit of the stability requirement, and the primary electric energy meter is taken as an example here. Then with local data segmentsThe first data point in the data series starts to traverse, the absolute value of the difference between two adjacent data points is calculated, when the absolute value of the difference is larger than the difference threshold value, the timestamp of the first data point is recorded as the fluctuation starting time point, then the traversing is continued until the absolute value of the difference is smaller than or equal to the difference threshold value, the timestamp of the last data point corresponding to the last data point larger than the difference threshold value is recorded as the fluctuation ending time point, the time interval between the fluctuation starting time point and the fluctuation ending time point is further obtained, and if the local data segment isIf there are multiple time intervals corresponding to the above successive time intervals greater than the difference threshold, the sum of all the time intervals is used as the local data segmentDuration of fluctuation in (a)。

Analyzing local data segmentsDuration of fluctuation in (a)In local data segmentsThe duty cycle in the corresponding duration to characterize the local data segmentThe larger the duty ratio is, the larger the data fluctuation is, and the condition that high equipment is suddenly switched in or switched out is met, thus, according to the local data segmentIs of the duration of fluctuation of (a)Obtaining a distance optimization factor of the ith data point according to the entropy value, the range and the mean value, wherein the calculation expression of the distance optimization factor of the ith data point is as follows:

Wherein, Representing the distance optimization factor of the ith data point, exp () represents an exponential function based on a natural constant, max () represents a maximum function, min () represents a minimum function, mean () represents a mean function,A local data segment representing the ith data point,Representing the duration of the fluctuation in the local data segment of the i-th data point,Representing the duration corresponding to the local data segment of the i-th data point,Representing the entropy value of the local data segment for the i-th data point.

It should be noted that, according to whether the probability distribution of the data points in the local data segment of the ith data point is uniform or not, the magnitude of the entropy value is determined, when the probabilities of q levels are completely equal, the probability is 1/p, and the entropy value is at mostConversely, when only one class exists, the probability of the class exists is 1, and when the rest is 0, the entropy value reaches the minimum of 0. Therefore, in the distance measurement, the distance optimization factor provided by the data points for realizing high load mutation is high, and the data points are more prone to be clustered and divided by using the time sequence distance; the fluctuation data of the unstable electric energy meter provides a distance optimization factor smaller than the high load mutation, and the distance optimization factor of the stable data is minimum, so that the high load mutation is more easily separated into the same cluster when the clustering distance between data points is calculated. The larger the entropy value, the larger the fluctuation of the local data segment of the ith data point, the more consistent with the characteristic of high load abrupt change, and the larger the distance optimization factor corresponding to the ith data point.

Similarly, obtaining a distance optimization factor of the jth data point, and then re-optimizing the clustering optimization distance between the two data points according to the distance optimization factor of the two data points, so as to correspondingly obtain a final clustering distance between the two data points, wherein the method comprises the following steps:

And taking the multiplication result of the distance optimization factor of the two data points and the clustering optimization distance between the two data points as the final clustering distance between the two data points.

In one embodiment, the calculation expression of the final cluster distance between the ith data point and the jth data point is:

It should be noted that, the clustering distance between the ith data point and the jth data point is further optimized by using the calculation expression of the final clustering distance, and the distance optimization factors obtained by combining the local fluctuation characteristics of the data points are combined according to the time sequence influence factors obtained by the neighbor window of the data points, so that in the clustering calculation of the current time sequence data of the electric energy meter, the data fluctuation caused by high load mutation can be classified, the distance measurement can be further scaled by the distance optimization factors obtained by the fluctuation characteristics, the data of the high load mutation and the unstable fluctuation of part of the electric energy meter can be distinguished, the problem that the data is not screened for the subsequent electric energy meter stability evaluation because the simple time sequence similarity is at the edge of the high load fluctuation cluster is solved.

So far, according to the acquisition method of the final clustering distance between the ith data point and the jth data point, the final clustering distance between every two data points in the current time sequence data is acquired respectively.

Step S104, based on the final clustering distance between every two data points in the current time sequence data, clustering all data points in the current time sequence data to obtain a corresponding clustering result, and removing the high-load fluctuation clustering cluster from the clustering result to obtain a final clustering result.

After the final clustering distance between every two data points in the current time sequence data is obtained, all data points in the current time sequence data can be clustered based on the final clustering distance between every two data points in the current time sequence data to obtain corresponding clustering junctions, specifically: and acquiring a K value when K-means clustering is carried out on all data points in the current time sequence data through an elbow rule, and carrying out K-means clustering on all data points in the current time sequence data based on the K value and the final clustering distance between every two data points in the current time sequence data to obtain a corresponding clustering result.

It should be noted that the elbow rule and the K-means cluster belong to the prior art, and are not described in detail here.

After the clustering of the current time sequence data is completed, removing the high-load fluctuation cluster from the clustering result to remove the stability evaluation error of the high-load fluctuation noise, and removing the high-load fluctuation cluster from the clustering result to obtain a final clustering result, wherein the method comprises the following steps of:

according to the data average value of each cluster in the clustering result, obtaining the average value of the data average values of all clusters, and determining that the clusters are high-load fluctuation clusters according to any cluster in the clustering result if the data average value of the clusters is larger than the average value of a first preset proportion or smaller than the average value of a second preset proportion;

Traversing each cluster in the clustering result to obtain all high-load fluctuation clusters, and removing all the high-load fluctuation clusters in the clustering result to obtain a final clustering result.

Preferably, the first preset ratio has a value range of [1.5,2], and the second preset ratio has a value range of [0,0.5], which can be adjusted according to the detected point electric energy meter. It is worth to say that the data average value of any cluster is larger than the average value of the first preset proportion, which indicates that the data points of the cluster belong to the condition of high load access, and the data average value of any cluster is smaller than the average value of the second preset proportion, which indicates that the data points of the cluster belong to the condition of high load disconnection.

So far, the high-load fluctuation clusters in the clustering result are removed, so that the stability of the subsequent electric energy meter is evaluated according to the reserved clusters.

Step S105, determining a stability evaluation result of the electric energy meter in a preset period according to the cluster discrete features in the final cluster result.

Carrying out data discrete condition analysis on each cluster in the final clustering result to determine a stability evaluation result of the electric energy meter in the preset period according to the cluster discrete characteristics in the final clustering result, wherein the stability evaluation result specifically comprises:

Respectively acquiring the data range and the data standard deviation of each cluster in the final clustering result to obtain the maximum data range and the maximum data standard deviation;

and if the maximum data range is smaller than or equal to the standard data range and the maximum data standard deviation is smaller than or equal to the standard data standard deviation, determining that the electric energy meter is in a stable running state in the preset period.

It should be noted that, the calculation of the polar error and the standard deviation belongs to the prior art, and will not be described in detail here. And secondly, after the discrete conditions of the data points of different clusters are acquired, comparing according to the stability requirement of the electric energy meter of the acquired data. According to the related regulations in national standard GB/T17215.322 electric energy meter accuracy requirement and test method, limit values of basic errors are regulated for electric energy meters of different types and grades, so that the maximum difference and the maximum standard deviation in the final clustering result can be compared with corresponding error limit values, and when the maximum difference and the maximum standard deviation meet the error limit values, the electric energy meter is determined to be in a stable operation state within a preset period, otherwise, the electric energy meter is determined to be in an unstable operation state.

In summary, in the operation process of the electric energy meter in the embodiment of the invention, current time sequence data in a preset period is collected, and a local data segment of any data point in the current time sequence data is obtained by taking the data point as the center; for any two data points in the current time sequence data, according to the similarity between the local data segments of the two data points, carrying out initial optimization on the clustering distance corresponding to the two data points in the clustering process, and correspondingly obtaining the clustering optimization distance between the two data points; for any one of the two data points, acquiring a distance optimization factor of the data point according to the data fluctuation of the local data segment of the data point, and re-optimizing the clustering optimization distance between the two data points according to the distance optimization factor of the two data points to correspondingly obtain the final clustering distance between the two data points; based on the final clustering distance between every two data points in the current time sequence data, clustering all data points in the current time sequence data to obtain a corresponding clustering result, and removing a high-load fluctuation clustering cluster from the clustering result to obtain a final clustering result; and determining a stability evaluation result of the electric energy meter in the preset period according to the clustering cluster discrete features in the final clustering result. The method comprises the steps of obtaining local data segments of each data point in current time sequence data, adding time sequence constraint when the current time sequence data are clustered, initially optimizing clustering distance among data points according to the local time sequence data among the data points, and meanwhile considering that partial fluctuation data mode features of an unstable electric energy meter are similar to high-load abrupt change data mode features in the clustering process, data under the conditions are easily clustered into one type, so that the clustering distance after the initial optimization of the added time sequence constraint is optimized again through the local fluctuation features of the local data segments of each data point in the current time sequence data, the clustering distance between any two data points in the clustering process is more rigorous, the high-load abrupt change data is prevented from being divided as abnormal data, and accuracy of stability evaluation results of the electric energy meter according to clustering results is improved.

The above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention.

Claims (6)

1. The stability evaluation method for the operation process of the electric energy meter is characterized by comprising the following steps of:

during the operation process of the electric energy meter, collecting current time sequence data in a preset period, and taking any data point in the current time sequence data as a center to obtain a local data segment of the data point;

for any two data points in the current time sequence data, according to the similarity between the local data segments of the two data points, carrying out initial optimization on the clustering distance corresponding to the two data points in the clustering process, and correspondingly obtaining the clustering optimization distance between the two data points;

for any one of the two data points, acquiring a distance optimization factor of the data point according to the data fluctuation of the local data segment of the data point, and re-optimizing the clustering optimization distance between the two data points according to the distance optimization factor of the two data points to correspondingly obtain the final clustering distance between the two data points;

Based on the final clustering distance between every two data points in the current time sequence data, clustering all data points in the current time sequence data to obtain a corresponding clustering result, and removing a high-load fluctuation clustering cluster from the clustering result to obtain a final clustering result;

Determining a stability evaluation result of the electric energy meter in the preset period according to the cluster discrete features in the final cluster result;

removing the high-load fluctuation cluster from the cluster result to obtain a final cluster result, wherein the method comprises the following steps:

according to the data average value of each cluster in the clustering result, obtaining the average value of the data average values of all clusters, and determining that the clusters are high-load fluctuation clusters according to any cluster in the clustering result if the data average value of the clusters is larger than the average value of a first preset proportion or smaller than the average value of a second preset proportion;

Traversing each cluster in the clustering result to obtain all high-load fluctuation clusters, and removing all the high-load fluctuation clusters in the clustering result to obtain a final clustering result.

2. The method for evaluating the stability of an operation process of an electric energy meter according to claim 1, wherein the performing initial optimization on the clustering distance corresponding to the two data points in the clustering process according to the similarity between the local data segments of the two data points, and correspondingly obtaining the clustering optimization distance between the two data points includes:

Acquiring a DTW distance between local data segments of the two data points, and carrying out normalization processing on the DTW distance, wherein an obtained normalization processing result is used as a time sequence influence factor;

And obtaining the corresponding clustering distance of the two data points in the clustering process, obtaining the addition result between a constant 1 and the time sequence influence factor, and taking the product between the addition result and the clustering distance as the clustering optimization distance between the two data points.

3. The method for evaluating the stability of an operation process of an electric energy meter according to claim 1, wherein the obtaining the distance optimization factor of the data point according to the data fluctuation of the local data segment of the data point comprises:

Dividing the difference value between the maximum value and the minimum value in the local data segment of the data point into a preset number of sections according to the dividing result, taking the ratio between the number of the data points contained in each section and the total number of the data points in the local data segment as the probability of the corresponding section respectively, acquiring the corresponding entropy value according to the probability of each section, and acquiring the range and the data value average value in the local data segment of the data point to obtain the first ratio between the range and the data value average value;

Calculating the absolute value of a difference between two adjacent data points from the first data point of the local data section of the data point, recording a first time stamp of the first data point if the absolute value of the difference is larger than a preset difference threshold value, continuing traversing the data points in the local data section of the data point until the absolute value of the difference is smaller than or equal to the difference threshold value, recording a second time stamp of the absolute value of the difference which is larger than the difference threshold value for the last time, acquiring the time interval between the first time stamp and the second time stamp, and taking the addition result of all the time intervals as fluctuation duration time if at least two time intervals exist in the local data section;

And acquiring the duty ratio of the fluctuation duration in the duration corresponding to the local data segment of the data point, substituting the product among the entropy value, the first ratio and the duty ratio into an exponential function with a natural constant as a base, and taking the obtained function result as a distance optimization factor of the data point.

4. The method for evaluating the stability of an operation process of an electric energy meter according to claim 1, wherein the re-optimizing the cluster optimization distance between the two data points according to the distance optimization factor of the two data points, correspondingly obtaining the final cluster distance between the two data points, comprises:

And taking the multiplication result of the distance optimization factor of the two data points and the clustering optimization distance between the two data points as the final clustering distance between the two data points.

5. The method for evaluating the stability of an electric energy meter according to claim 1, wherein the determining the stability evaluation result of the electric energy meter within the preset period according to the cluster discrete features in the final cluster result comprises:

Respectively acquiring the data range and the data standard deviation of each cluster in the final clustering result to obtain the maximum data range and the maximum data standard deviation;

and if the maximum data range is smaller than or equal to the standard data range and the maximum data standard deviation is smaller than or equal to the standard data standard deviation, determining that the electric energy meter is in a stable running state in the preset period.

6. The method for evaluating the stability of an operation process of an electric energy meter according to claim 1, wherein the clustering all data points in the current time series data based on the final clustering distance between every two data points in the current time series data to obtain corresponding clustering results comprises:

And acquiring a K value when K-means clustering is carried out on all data points in the current time sequence data through an elbow rule, and carrying out K-means clustering on all data points in the current time sequence data based on the K value and the final clustering distance between every two data points in the current time sequence data to obtain a corresponding clustering result.